Almost every country on Earth now holds elections. Yet trust in democracy is falling almost everywhere.

The V-Dem Institute, which tracks political freedom in 179 countries, reports that the level of democracy enjoyed by the average person has fallen back to where it was in 1978. Autocracies now outnumber democracies, 92 to 87. About three quarters of the world's population lives under some form of autocracy (V-Dem Democracy Report 2026 - Unraveling The Democratic Era?).

There are many reasons for this, but one stands out. Elected governments struggle to deliver on the problems people care about most: education, jobs, housing, climate change. When democracy keeps failing to solve problems, strongmen who promise to solve them start to look attractive.

One major and underappreciated cause is that we hold a primitive view of democracy, and it delivers poorer representation than we could have.

Most people think that if we vote for one candidate every four or five years, the box is ticked and our government is democratic. But democracy is far richer and more nuanced than that. It includes tools we rarely use, such as citizens' assemblies chosen by lot (sortition) and referendums.

This article is about something even more basic: the ballot itself. Our ballots throw away almost everything voters could tell us. And the one reform that would fix this has been tried, judged a failure and set aside, for the wrong reasons.

In most countries, a ballot lets you say exactly one thing: which single candidate or party you prefer. It cannot say how much you like them, whether you could live with the runner-up, or whether you despise everyone on the list.

Some countries do better. Australia, for example, lets voters rank candidates in order of preference. But a ranking still hides the distance between choices. Ranking someone second could mean "almost as good as my favorite" or "slightly less terrible than the worst".

This is how we end up voting for the lesser evil. Consider an election with 100 voters and three candidates:

• 42 voters love Candidate A, hate Candidate B and quite like Candidate C.
• 38 voters love B, hate A and quite like C.
• 20 voters love C and dislike both A and B.

Under the usual rules, A wins with 42 votes, even though 58 of the 100 voters dislike A. Now suppose each voter scores every candidate from −10 (the worst) to +10 (the best):

Candidate	First choices	Average score
A	42	−0.6
B	38	−1.4
C	20	+6.0

The scores tell a completely different story. The candidate who wins under today's rules is, on balance, rejected by the electorate. The candidate who finishes last in first choices is the one almost everyone can accept. A one-choice ballot cannot see this. A scored ballot cannot miss it.

The fix is simple to describe. Instead of choosing one name, each voter scores every candidate on a scale. One possible range:

• +10 for a candidate you would be delighted to see elected
• 0 for a candidate you feel neutral about
• −10 for a candidate you would hate to see in office

The candidate with the best overall score wins.

Scoring works for any group decision, not just elections. Choosing where to have dinner with friends is the same problem: one friend loves sushi, another cannot stand it, and pizza is everyone's second choice. Ask each person to score the options and pizza wins, while a show of hands might have sent half the table home unhappy.

Voters do not have to score everyone. A blank counts as 0, so on the ballot or the voting machine every candidate starts at neutral, and you change only the ones you have an opinion about. Someone who wants to vote the old way can give one candidate +10 and leave the rest alone.

With a scored ballot, you no longer have to vote for the lesser of two evils. You can give your favorite small party +10, the tolerable major candidate +2 and the one you fear −10, all at once, without "wasting" your vote.

This idea is not new. Electoral reformers have argued for it for decades. Its simplest form, approval voting, lets voters tick every candidate they find acceptable, which amounts to a scale with only two points. A close cousin, majority judgment, asks voters to grade candidates with words such as "good" or "poor" and picks the one with the best median grade.

For a friendly introduction to these and other voting systems, Mr. Beat's video What's the best way to vote? is a good place to start.

No country has used a full −10 to +10 scale in a binding national election. But close relatives of score voting have been used and studied, and the record is mixed.

The Good

• Scored ballots favor candidates most people can accept. During the 2012 French presidential election, researchers set up booths next to real polling stations and asked 2,340 voters to fill in a second, experimental ballot grading every candidate. The usual two-round system favored candidates who provoked strong feelings; scored ballots favored candidates with support across a broad part of the electorate (Who's favored by evaluative voting? An experiment conducted during the 2012 French presidential election). A follow-up study found that the scale itself matters: offering negative grades changed the results, because a negative grade carries a symbolic power of rejection that hits some candidates far harder than others (How voters use grade scales in evaluative voting).
• People do say more when they are allowed to. Fargo, North Dakota, adopted approval voting in 2018. In its 2022 city commission race, with 15 candidates for two seats, voters approved about three candidates each (DemocracySOS).
• It has survived political attacks. St. Louis uses approval voting in its primaries. When Missouri banned ranked and approval voting statewide in 2024, St. Louis was exempted, and attempts to repeal its system have failed (Wikipedia).

The Bad

• Tactical voting is real. In Fargo's 2022 mayoral race, FairVote estimated that about 60% of voters approved only one of seven candidates. Some campaigns told supporters not to "dilute" their vote by approving rivals (DemocracySOS).
• Results were reported badly. Fargo's official results showed each winner's share of all approvals rather than the share of voters who approved them. Winners looked as if they had very little support.
• Some adopters gave up. The IEEE, a large international engineering association, used approval voting from 1987 and dropped it in 2002 because "few of our members were using it". Dartmouth's alumni association replaced it with runoff elections in 2009 after a controversy (Wikipedia).
• Score scales can be gamed. Olympic figure skating used panels of judges giving numerical scores until the 2002 Salt Lake City scandal, when a judge admitted favoring one pair. The reform that followed dropped the highest and lowest scores before averaging, making a single biased judge far less effective (CBC).

Look closely at those failures and a pattern appears. Every place that adopted a richer ballot asked voters to do more work, then threw away almost everything the ballots said. The scores were collected, a winner was picked, and the numbers were forgotten.

In other words, score voting has paid all of its costs and collected almost none of its benefits. Voters had to learn a new ballot. Election officials had to explain new results. Campaigns found new ways to game them. In return, voters got a slightly different way of choosing the same thing: one winner.

No wonder people were unimpressed. Judging score voting this way is like buying a car and only ever using its radio.

A scored ballot is not just a better way to count votes. It is a detailed measurement of how a whole society feels about every candidate, and that measurement can do many jobs at once.

Here is what the same set of scores could do if we stopped throwing it away.

1. Pick the Winner(s)

The candidate, or candidates, with the best scores win. This is the only use score voting has had so far.

2. Measure the Mandate

Today, a winner is a winner. Whether they were loved or barely tolerated, they claim a mandate from "the people". With scored ballots, everyone can see the difference. A president elected with an average of +7 has a broad mandate. One elected with +3 knows that most voters accepted them without enthusiasm, and so does everyone else. Results reported this way would also make it obvious when a society is growing more or less satisfied with its choices from one election to the next.

3. Reject Bad Candidates

Today, voters who dislike every candidate have no way to say so. They abstain or spoil their ballot, and someone wins anyway. With scored ballots, a simple rule becomes possible: if no candidate ends up with a positive score, the election is held again with entirely new candidates.

Versions of this already exist. In 2018, the state election commissions of Maharashtra and Haryana in India ruled that in local elections, "None of the Above" would be treated as a fictional candidate; if it received the most votes, fresh elections would be held (Scroll.in). Nevada has long offered "None of these candidates" on its ballots, but it can never win, so the protest is counted and then ignored. Russia offered an "against all" option until its parliament abolished it in 2006 (RFE/RL).

A scored ballot does better than any of these, because it captures rejection even from voters who also gave someone a positive mark. Three safeguards would keep it workable:

• Use a robust measure. In a deeply divided country, half the voters might give their side +10 and the other side −10, leaving everyone near zero. Requiring that a winner be scored positively by more than half of voters, or using the median score instead of the average, means a rerun happens only when most voters reject every candidate.
• Allow one rerun only. If the second election also fails, the best-scored candidate takes office, with the weak result published for all to see.
• Keep the government running. The incumbent or a caretaker stays in office until the rerun, as already happens while coalition governments are being formed.

The rule would rarely need to be used. Knowing it exists, parties would choose candidates acceptable beyond their own base.

The same scores can also deal with individual candidates. A candidate whose average score falls to −8 or below has been rejected by an overwhelming share of voters. Such a candidate could be barred from standing for public office for a period, say ten years.

This needs care. A majority could use it against a candidate from a minority it dislikes, so the threshold must be very high and the penalty limited to standing for election, never to other rights.

4. Fund New Ideas

Many countries fund political parties with public money according to their election results. Germany, for example, funds parties that win at least 0.5% of the vote in federal or European elections, or 1% in state elections, far below the 5% needed to win seats (State financing of parties).

But these systems count only first choices. If funding followed the scores instead, inclusive parties and parties with new ideas could grow between elections, even before they win a single seat. Counting only positive scores would also stop rival supporters from cutting a party's funding by scoring it −10.

5. Trigger a Recall

The same score system need not wait for election day. A government's citizen portal could let people score any elected official at any time. If an official's score stays very low for months, or if a large majority of registered voters, say 80%, call for their removal, a recall election is held, or in the most extreme cases the official is removed at once.

Counting registered voters rather than the people who happen to take part stops a small, motivated group from removing an official on a quiet day. And limiting this to elected officials keeps the civil service from becoming a popularity contest. Even so, the possibility of removal would give officials a reason to fear the consequences of bad behavior between elections, not just at them.

6. Read the Correlations

A score ballot is also one of the richest opinion surveys ever taken, covering the whole electorate. Statisticians could publish how voters' scores for different parties relate to each other, using anonymous totals only, so no individual ballot is ever revealed.

For example, how closely do voters' scores for a Labour party and a Green party move together? If people who score one highly almost always score the other highly too, the two parties are competing for the same supporters, and each has a reason to show what really sets it apart. If the scores barely move together, they are speaking to different people, and could cooperate without losing voters. Correlations with age, region or income, already common in exit polls, would become far more precise.

This would help parties understand what really matters to voters, and help voters see the political landscape as it is rather than as party slogans describe it.

A richer ballot only helps if people use it, and if it is hard to game. A few simple measures would go a long way.

• A Democracy Day. If scores decide mandates, funding and recalls, everyone must be able to vote. Election day should be a public holiday, as it already is in South Korea, or fall on a weekend, as in Australia, Brazil and much of Europe.
• A scoring rule that resists tactics. Some voters will always give their favorite +10 and everyone else −10. Figure skating shows one answer: drop the most extreme scores before averaging. The median score, used by majority judgment, is another. Both make exaggeration far less effective than with a plain average.
• Neutral by default. Starting every candidate at 0 means nobody is forced to score people they know nothing about, and a voter who scores only one candidate is no worse off than today.
• Clear results. Fargo's experience shows that badly presented results can undermine a good system. Results should show each candidate's average or median score and the share of voters who scored them positively, among other useful statistics, in plain numbers anyone can read.

Any voting method needs a little public education, because some of the habits we learned under one-choice voting no longer apply, and a few new myths are bound to spread. Voters should hear plainly what is true and what is not. For score voting in particular:

• Myth: "Scoring a backup candidate is a betrayal of my favorite."

  Not so. You can give your favorite +10 and your backup +7 on the same ballot. Under one-choice voting, supporting a backup meant abandoning your favorite. Here you can support both, each as much as you really like them.

• Myth: "A high score for my backup can never hurt my favorite."

  This one is also false, and it matters. If your favorite and your backup are the two front-runners, the points you give your backup help it compete with your favorite. That is not a flaw: it is the ballot doing exactly what you told it to. So score carefully. If you give your favorite +10 and your backup +8, and then feel disappointed when the backup wins, your scores were not honest. The backup was never really a +8 for you. Perhaps it was a +4.

  A simple test: imagine each candidate actually winning, and ask how you would feel. The gap between your scores should match the gap between those feelings.

• Myth: "Scoring a small party is a wasted vote."

  No. Your score for a small party sits alongside your scores for everyone else. It costs your other choices nothing, and it contributes to the party's reputation, and perhaps to its public funding, even if the party wins no seat.

• Myth: "Leaving a candidate blank hurts them."

  A blank counts as 0, neutral. It neither helps nor hurts. Give a negative score only when you actually want a candidate kept out of office.

• Myth: "Giving everyone else −10 is the smart move."

  It is a blunt move. It throws away your chance to separate the candidate you could live with from the one you fear, which is the very information the rerun, funding and disqualification rules depend on. And with a median or trimmed scoring rule, exaggeration gains you little.

A short, neutral guide like this could be printed on every ballot paper, shown on the voting machine and repeated in public campaigns before the first election.

"Most countries elect parliaments, not presidents."

True, and plain score voting is built for single winners. Filling a whole parliament this way would help broadly acceptable candidates of every background, but a group whose views the majority dislikes could be shut out entirely: if 60% of voters score their side +10 and the other side −10, the 60% could take every seat, while proportional representation would give the 40% four seats in ten. Proportional versions of score voting have been proposed, and countries with proportional parliaments do not have to choose: they can keep their method for allocating seats and still collect scores for everything else in this article, from mandates and funding to rejection and correlations.

"Detailed ballots make vote buying easier."

With a range of scores for each of many candidates, a voter could mark a unique pattern on minor candidates, and someone paying for votes could later look for that pattern in the published ballots. The answer is simple: publish only totals and anonymous statistics, never individual ballot records.

"An unknown candidate could win by accident."

If only a handful of people score an obscure candidate, and they all give +10, that candidate's average among those voters is perfect. That is why blanks must count as 0 across the whole electorate, and why a winner should also be scored positively by a minimum share of all voters. A few enthusiastic fans should never outweigh a whole country's indifference.

"Who chooses the thresholds?"

The numbers in this article, −8 for a ban, 80% for a recall, ten years out of office, are illustrations, not a final design. Whatever numbers a country chooses, they must be beyond the reach of the politicians they constrain: written into the constitution or set by an independent body, and changeable only by referendum.

"How could this ever happen?"

The politicians who would have to adopt a new voting system were elected under the old one, and rarely want to change the rules that got them there. But change can start small. Fargo and St. Louis adopted approval voting through citizen ballot initiatives, and Oregon's Independent Party used STAR voting, a close relative of score voting, in its 2020 primary. Clubs, unions, universities, professional associations and party primaries can all try it first.

And we can start even smaller. Score the options the next time your team at work has to choose a project, or your friends have to choose where to have dinner or which movie to watch. Small, everyday decisions build confidence in the method, show people that it works, and let best practices evolve long before anything as important as a national election depends on them.

The same thinking applies beyond electing people.

• Scored referendums. A yes-or-no referendum forces a complex question into two answers, and often neither side's preferred outcome is what most people want. Letting voters score several options would reveal the option most people can live with.
• A name on every law. When a law passes, one of the legislators who voted for it could be picked at random and have their name attached to it. If the law works, they share the credit. If it produces unintended consequences, their reputation suffers. Votes are already public, but a name on the law makes responsibility visible to everyone, and gives lawmakers a reason to read what they vote for.

Democracy is not failing because people stopped caring. It is failing partly because we give them a ballot that can barely hear them.

Score voting was never comprehensively tested. It was asked to do one job, picking a winner, and judged on the effort it cost. Used fully, the same ballot can show how strong a mandate is, reject bad candidates, fund the ideas of tomorrow, keep the most despised out of office, hold officials to account between elections and reveal how voters really see the parties.

No voting system is perfect. Every one of them can be gamed or misunderstood in some way. But perhaps the best principle is to start with the system that lets people express their true feelings as precisely as possible. Once that information is collected, mechanisms can be added to reduce gaming and misunderstanding, such as robust scoring rules, clear results and public education. A ballot that hears too little cannot be fixed later; a ballot that hears a lot can always be refined.

The information is already there, in every voter's head. All we have to do is stop throwing it away.